Comparative performance analysis of quantum feature maps for quantum kernel-based machine learning

Area of Science:

• Quantum Computing
• Machine Learning

Background:

• Quantum algorithms offer enhanced solutions for machine learning.
• Quantum kernels are a key area for achieving these enhancements.

Purpose of the Study:

• Analyze quantum kernel approaches using feature maps and hyperparameters.
• Develop enhanced quantum kernels for improved machine learning performance.

Main Methods:

• Introduced a novel high-order feature map and evaluated five existing state-of-the-art feature maps.
• Assessed the impact of the rotational factor hyperparameter on kernel performance.
• Examined hyperparameter-tuned feature maps for enhanced decision boundaries and kernel expressivity.

Main Results:

• A new high-order feature map was developed and tested alongside five existing feature maps.
• The rotational factor was identified as a significant hyperparameter for kernel performance improvement.
• Analysis on nonlinear datasets demonstrated that optimized quantum feature maps enhance decision boundaries.

Conclusions:

• Well-tuned quantum feature maps significantly improve the generalization ability of quantum kernels.
• This research advances the effectiveness of quantum kernels for diverse quantum-enhanced machine learning applications.